Dynamic DNS

HIBBS, BARR (PB) RBHIBBS at msg.pacbell.com
Fri Aug 6 18:48:06 UTC 1999


Brian and Irina--

tricky stuff, huh?  Badly complicated by roaming....

Rather than comment on your development, let me suggest some criteria we
should use for evaluating approaches and proposed solutions.

I have a lot (thousands!) of users who roam from one subnetwork to another,
and even from one docking station to another on the same subnet.  Their
"home" docking station is configured with a static address (with static A
and PTR records as well), and when they roam "normal" practice is to use a
PCMCIA Ethernet adapter that is plugged-in to any free port where they
receive a dynamic address.

I have not yet implemented dynamic DNS updates for any of my clients, but if
it were activated this is the point when I would expect that the update
would occur.

A roaming user could also dock their laptop in an available docking station
and begin to use the static address associated with that device (our docking
stations have NICs).  This is the case where we have telecommute "hotels"
set-up for our roaming users.  A static A and PTR record here is a major
annoyance because our proxy servers require a PTR record for the specific
host accessing certain off-network services, and Microsoft clients don't
like to be told their hostname (option 12).  Ideally, hotel docking stations
would perform dynamic DNS update even though their IP address assignments
are static.

I can also pretty much guarantee that no roaming user will ever explicitly
release an address before un-docking or un-plugging and moving on, so that
the dynamic update strategy must account for this in some way.

Now, with regard to the MX and other RR types....  I basically do not want
the DHCP server to become a surrogate administrator of DNS, so before
embarking on development to add additional RR types to the DHCP-DNS update,
I really would like to see that worked into Mark Stapp's current draft of
the interaction protocol so that all the interested parties can review and
comment.

I also am much less sensitive to the effects of roaming on mail delivery
because none of our users currently have MX records pointing to anywhere
besides their assigned mail server.

--Barr





More information about the dhcp-hackers mailing list